Readme for replication package to reproduce the results of 

Work from Home & Productivity: Evidence from Personnel & Analytics Data on IT Professionals

by Gibbs, Mengel, Siemroth

Date: 19.4.2022

The replication package contains 
* a copy of the programs (Stata do-file) used to create the final results.
* this read-me file including details of how the proprietary data can be obtained by others.

Stata version used: 16.0

*** Description of Stata-do-file ***

- reproduction.do does all the data preparation and analysis based on the source data files
- the first part prepares the data and saves the dataset as "productivitydataset.dta"
- the second part produces graphs/tables of all results in Sections 3.1 and 3.2 as well as corresponding Appendix Tables.
- the third part produces graphs/tables of all results in Sections 3.3 as well as corresponding Appendix Tables. The third part uses "WFH_WPA.dta" which is as received by the company as well as a data set that merges "productivitydataset.dta" and "WFH_WPA.dta". The do-file loads the correct data set for each table/graph.
- the do file indicates which table/figure is being produced (numbering based on pre-print PDF)
- Stata variables mostly match the variable names used in the paper. One exception: "KPIValue" in Stata is "Output" in the paper. Sometimes
variable names are not exactly the same but extremely similar (e.g. "all_emails" in Stata is "Emails" in the paper). Any queries can be addressed to the authors.
- It took about 5 min to run the first and second part (i5-6200U CPU, 8GB RAM, SSD) and about 3 min to run the third part.

*** Description of Sourcefiles ***

- 04.2019-08.2020.demographics.salary.performance.xlsx contains the Sapience and demographic variables (without commute time)
- It is on employee-month level in a long format (i.e., each row is an employee-month)
- CommuteTime_Professors_v1.0.xlsx contains the estimate of the commute time for the employees
- 01.2020-08.2020.WPA.xlsx contains the WPA variables of the smaller WPA sample (see the paper)
- It is on employee-week level in a long format (i.e., each row is an employee-week)

*** Requests for Data Access ***

The data are confidential as they include personal details about individual employees. 
Researchers interested in gaining access to the data should contact Michael Gibbs, gibbs@uchicago.edu. 
Please provide a description of the project and intended use of the data. 
Gibbs will contact the firm, which will evaluate requests. 
If approved the researchers will have to sign a non-disclosure agreement with the company, and take proper procedures to ensure data security.
